CC14 has been shown previously to be metabolized to the trichloromethyl radical (.CC13) and to a novel oxygen-containing carbon dioxide anion radical (.C02-) in the perfused rat liver. These free radicals were detected by electron spin resonance using the spin-trapping technique. The .C02- radical adduct also was observed in urine following the intragastric administration of CC14 or CBrC13 and spin trap. Detection of the .C02- adduct in the effluent perfusate was decreased 3-4 fold by DIDS (0.2 mM), an inhibitor of the plasma membrane anion transport system. The rate of formation of .CO2- radical adduct was decreased 2-3 fold following inhibition of cytochrome P-450-dependent mono-oxygenases by metyrapone (0.5 mM) and was increased about two-fold by induction of cytochrome P-450 by phenobarbital pretreatment. Toxicity of halocarbons in the perfused liver was assessed by measuring the release of lactate dehydrogenase (LDH) into the effluent perfusate in livers from phenobarbital-treated rats under conditions identical to those employed to detect radical adducts (i.e., during the infusion of CC14 or CBrC13 into livers perfused with either nitrogen- or oxygen-saturated perfusate). Metabolism of halocarbons to .C02- radical adduct was 6-8 fold faster during perfusion with nitrogen-saturated rather than with oxygen-saturated perfusate. Concomitantly, liver damage detected from LDH release occurred much sooner during halocarbon infusion in the presence of nitrogen-saturated perfusate. A good correlation (r= -0.80) between the rate of formation of PBN/.C02- and the time to onset of LDH release following halocarbon infusion was observed. Therefore, it is concluded that PBN/.C02- is a useful marker for oxygen-containing free radical intermediates which may be causally related to halocarbon-induced hepatotoxicity. Recently, the .CC13 and .C02-radical adducts also have been detected in the bile from anesthetized rats.